Is AI the answer to maintaining the integrity of remote exams?
As online spaces become increasingly integral to our lives, virtual assessments and interviews have become the norm. This surge in popularity is accompanied by a corresponding rise in the demand for remote proctoring solutions to safeguard the integrity of these online evaluations, prompting the development of a variety of innovative, artificial intelligence (AI)-powered solutions.
These solutions offer unique advantages in the ever-evolving landscape of online proctoring. Available in three distinct modes — live (which combines the efficiency of AI with human judgement), auto (which operates entirely autonomously, requiring no human proctor presence), and record and review (which records the exam session for later review by human proctors) — they establish a secure, efficient and inclusive online learning environment.
What the early days of AI remote proctoring were like
The early AI-powered remote proctoring systems held immense promise for streamlining online assessments, but their limitations raised significant concerns. They included:
- Limited capabilities: The systems were vulnerable to simple tactics, with masks and pre-recorded footage easily bypassing the systems; rigid algorithms raised a high number of false positives; and, blind spots in the systems were unable to detect sophisticated cheating techniques, such as switching browser tabs or seeking external help.
- Negative user experience: Constant scrutiny and the high potential for false positives heightened stress and anxiety among students; the systems were not flexible enough to accommodate natural behavioural variations such as repetitive movements or impulsivity; and, the systems were not friendly to technologically-challenged and differently-abled participants.
These initial challenges highlighted the need for further development and refinement of AI proctoring systems.
Today’s AI proctors
Moving beyond their initial limitations, modern AI proctoring systems include the following features:
- Advanced behaviour analysis: The systems can detect subtle behavioural cues, including browsing of unauthorised tabs, unauthorised device usage or unusual eye movements.
- Comprehensive audio monitoring: AI algorithms are now trained to detect attempted communication with external sources during an exam. Whispering or muffled conversations and unusual background noise are flagged for further investigation.
- Environment monitoring: The technology can now identify unauthorised individuals entering the exam room.
As it happens, the advancements have not solely focused on exam integrity; inclusivity has also become a key consideration. They include:
- Features catering to persons with disabilities: Screen readers, text-to-speech conversion and adjustable font sizes allow persons with diverse needs to participate in online assessments on an equal footing.
- Accessibility improvements: More intuitive and user-friendly user interfaces ensure that people with varying technical skills can navigate the platform comfortably and focus on their exams.
This dual focus on security and inclusivity marks a significant step forward for AI proctoring.
What the future holds
The future of AI proctoring is brimming with potential and promise. As technology continues to evolve and mature, we can anticipate significant advancements:
- Enhanced integrity, reduced false positives: Future algorithms promise to be more sophisticated, catching complex cheating methods while minimising false alarms.
- Seamless integration with online learning platforms: Likely integration with learning platforms simplifies exam administration for administrators and participants and fosters a holistic online learning experience.
- A shift towards more natural systems: Systems adapt by learning from student behaviour and identify situations that necessitate human judgement and alert accordingly.
- Role of human proctors redefined: Human proctors transition to supervisory roles, focussing on investigating flagged activities and acquiring new skill sets like data analysis and communication for effective collaboration with AI.
- Navigation of potential challenges: Moving forward, a balanced approach is crucial, that is, embracing AI while preserving the human element in recruitment. Additionally, maintaining data confidentiality and security is paramount.
- Vision of a brighter future: AI proctoring shines with efficiency (reduced workload for human proctors) leading to cost savings and scalability in exam administration. Plus, accessibility improvements broaden its reach.
With its potential for security, fairness and accessibility, AI proctoring holds the key to a future that includes successful online learning for everyone.
Conclusion
From its rapid ascent to cutting-edge solutions, the evolution of AI proctoring has been impressive. Looking ahead, AI algorithms are poised to attain greater complexity, emulating human proctoring with increasing accuracy. But balance is key: embrace AI’s potential while prioritising human oversight and data security. By working together, AI and humans can unlock a future that will enrich online learning for all.
